diff --git a/kernels/armsve/3/armsve_asm_2vx10.h b/kernels/armsve/3/armsve_asm_2vx10.h
index 8e37585cb..ae89fa1ec 100644
--- a/kernels/armsve/3/armsve_asm_2vx10.h
+++ b/kernels/armsve/3/armsve_asm_2vx10.h
@@ -130,6 +130,13 @@
   SCALE_COL4(Z12,Z13,Z14,Z15,ZFACTOR) \
   SCALE_COL4(Z16,Z17,Z18,Z19,ZFACTOR)
 
+#define GEMM_C_FMLA_UKER(C0FH,C1FH,C2FH,C3FH,C4FH,C0LH,C1LH,C2LH,C3LH,C4LH,PT,Z0FH,Z1FH,Z2FH,Z3FH,Z4FH,Z0LH,Z1LH,Z2LH,Z3LH,Z4LH,ZSCALE) \
+  GEMM_FMLA2(C0FH,C0LH,PT,Z0FH,Z0LH,ZSCALE) \
+  GEMM_FMLA2(C1FH,C1LH,PT,Z1FH,Z1LH,ZSCALE) \
+  GEMM_FMLA2(C2FH,C2LH,PT,Z2FH,Z2LH,ZSCALE) \
+  GEMM_FMLA2(C3FH,C3LH,PT,Z3FH,Z3LH,ZSCALE) \
+  GEMM_FMLA2(C4FH,C4LH,PT,Z4FH,Z4LH,ZSCALE)
+
 #define GEMM_C_FMAD_UKER(Z0FH,Z1FH,Z2FH,Z3FH,Z4FH,Z0LH,Z1LH,Z2LH,Z3LH,Z4LH,PFH,PLH,C0FH,C1FH,C2FH,C3FH,C4FH,C0LH,C1LH,C2LH,C3LH,C4LH,ZSCALE) \
   GEMM_CCOL_FMAD(Z0FH,Z0LH,PFH,PLH,C0FH,C0LH,ZSCALE) \
   GEMM_CCOL_FMAD(Z1FH,Z1LH,PFH,PLH,C1FH,C1LH,ZSCALE) \
diff --git a/kernels/armsve/3/bli_gemm_armsve_asm_d2vx10_unindexed.c b/kernels/armsve/3/bli_gemm_armsve_asm_d2vx10_unindexed.c
index b48117ce0..e5b78a592 100644
--- a/kernels/armsve/3/bli_gemm_armsve_asm_d2vx10_unindexed.c
+++ b/kernels/armsve/3/bli_gemm_armsve_asm_d2vx10_unindexed.c
@@ -264,20 +264,17 @@ SCALE_COL20(z0,z1,z2,z3,z4,z5,z6,z7,z8,z9,z10,z11,z12,z13,z14,z15,z16,z17,z18,z1
 "                                                 \n\t"
 " WRITE_MEM_C:                                    \n\t" // Available scratch: Z[20-30].
 "                                                 \n\t" // Here used scratch: Z[20-29].
-" fmov            s28, #0.0                       \n\t"
-" fmov            w16, s28                        \n\t"
-" cmp             w16, w8                         \n\t"
+" fcmp            d31, #0.0                       \n\t" // Skip loading if *beta == 0 to override NaN.
 " b.eq            BETA_ZERO_C                     \n\t"
-"                                                 \n\t"
 // First half of C is already loaded in this case.
-GEMM_C_FMAD_LOAD_UKER_C(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,p0,p0,z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,z31,x9,x7)
-GEMM_C_FMAD_UKER(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,p0,z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,z31)
+// GEMM_C_FMAD_LOAD_UKER_C(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,p0,p0,z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,z31,x9,x7)
+GEMM_C_FMLA_UKER(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z31)
+GEMM_C_LOAD_UKER_C(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,p0,p0,x9,x7)
+GEMM_C_FMLA_UKER(z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,p0,z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z31)
 "                                                 \n\t"
 " BETA_ZERO_C:                                    \n\t"
-"                                                 \n\t"
-GEMM_C_STORE_UKER_C(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,p0,p0,x5,x7)
 GEMM_C_STORE_UKER_C(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,p0,x5,x7)
-"                                                 \n\t"
+GEMM_C_STORE_UKER_C(z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,p0,p0,x5,x7)
 " b               END_WRITE_MEM                   \n\t"
 "                                                 \n\t"
 " WRITE_MEM_G:                                    \n\t" // Available scratch: Z[20-30].
@@ -286,20 +283,18 @@ GEMM_C_STORE_UKER_C(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,p0,x5,x7)
 " incb            x8                              \n\t"
 " madd            x8, x8, x6, xzr                 \n\t" // C-column's logical 1-vector skip.
 " index           z30.d, xzr, x6                  \n\t" // Skips passed to index is not multiplied by 8.
-" fmov            s28, #0.0                       \n\t"
-" fmov            w16, s28                        \n\t"
-" cmp             w16, w8                         \n\t"
+"                                                 \n\t"
+" fcmp            d31, #0.0                       \n\t" // Skip loading if *beta == 0 to override NaN.
 " b.eq            BETA_ZERO_G                     \n\t"
 "                                                 \n\t"
 GEMM_C_LOAD_UKER_G(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z30,p0,p0,x9,x7,x8,x16)
-GEMM_C_FMAD_UKER(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,p0,p0,z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,z31)
-GEMM_C_LOAD_UKER_G(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,z30,p0,p0,x9,x7,x8,x16)
-GEMM_C_FMAD_UKER(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,p0,z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,z31)
+GEMM_C_FMLA_UKER(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z31)
+GEMM_C_LOAD_UKER_G(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z30,p0,p0,x9,x7,x8,x16)
+GEMM_C_FMLA_UKER(z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,p0,z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z31)
 "                                                 \n\t"
 " BETA_ZERO_G:                                    \n\t"
-"                                                 \n\t"
-GEMM_C_STORE_UKER_G(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z30,p0,p0,x5,x7,x8,x16)
 GEMM_C_STORE_UKER_G(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,z30,p0,p0,x5,x7,x8,x16)
+GEMM_C_STORE_UKER_G(z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,z30,p0,p0,x5,x7,x8,x16)
 "                                                 \n\t"
 " END_WRITE_MEM:                                  \n\t"
 " b               END_EXEC                        \n\t"
diff --git a/kernels/armsve/3/bli_gemm_armsve_asm_s2vx10_unindexed.c b/kernels/armsve/3/bli_gemm_armsve_asm_s2vx10_unindexed.c
index 94bc08ad9..00b3f20b4 100644
--- a/kernels/armsve/3/bli_gemm_armsve_asm_s2vx10_unindexed.c
+++ b/kernels/armsve/3/bli_gemm_armsve_asm_s2vx10_unindexed.c
@@ -252,20 +252,16 @@ SCALE_COL20(z0,z1,z2,z3,z4,z5,z6,z7,z8,z9,z10,z11,z12,z13,z14,z15,z16,z17,z18,z1
 "                                                 \n\t"
 " WRITE_MEM_C:                                    \n\t" // Available scratch: Z[20-30].
 "                                                 \n\t" // Here used scratch: Z[20-29].
-" fmov            s28, #0.0                       \n\t"
-" fmov            w16, s28                        \n\t"
-" cmp             w16, w8                         \n\t"
+" fcmp            s31, #0.0                       \n\t"
 " b.eq            BETA_ZERO_C                     \n\t"
-"                                                 \n\t"
 GEMM_C_LOAD_UKER_C(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,p0,p0,x9,x7)
-GEMM_C_FMAD_UKER(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,p0,p0,z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,z31)
-GEMM_C_LOAD_UKER_C(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,p0,x9,x7)
-GEMM_C_FMAD_UKER(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,p0,z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,z31)
+GEMM_C_FMLA_UKER(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z31)
+GEMM_C_LOAD_UKER_C(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,p0,p0,x9,x7)
+GEMM_C_FMLA_UKER(z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,p0,z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z31)
 "                                                 \n\t"
 " BETA_ZERO_C:                                    \n\t"
-"                                                 \n\t"
-GEMM_C_STORE_UKER_C(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,p0,p0,x5,x7)
 GEMM_C_STORE_UKER_C(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,p0,x5,x7)
+GEMM_C_STORE_UKER_C(z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,p0,p0,x5,x7)
 " b               END_WRITE_MEM                   \n\t"
 "                                                 \n\t"
 " WRITE_MEM_G:                                    \n\t" // Available scratch: Z[20-30].
@@ -274,20 +270,17 @@ GEMM_C_STORE_UKER_C(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,p0,x5,x7)
 " incb            x8                              \n\t"
 " madd            x8, x8, x6, xzr                 \n\t" // C-column's logical 1-vector skip.
 " index           z30.s, wzr, w6                  \n\t" // Skips passed to index is not multiplied by 8.
-" fmov            s28, #0.0                       \n\t"
-" fmov            w16, s28                        \n\t"
-" cmp             w16, w8                         \n\t"
-" b.eq            BETA_ZERO_G                     \n\t"
 "                                                 \n\t"
+" fcmp            s31, #0.0                       \n\t"
+" b.eq            BETA_ZERO_G                     \n\t"
 GEMM_C_LOAD_UKER_G(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z30,p0,p0,x9,x7,x8,x16)
-GEMM_C_FMAD_UKER(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,p0,p0,z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,z31)
-GEMM_C_LOAD_UKER_G(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,z30,p0,p0,x9,x7,x8,x16)
-GEMM_C_FMAD_UKER(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,p0,z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,z31)
+GEMM_C_FMLA_UKER(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,p0,z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z31)
+GEMM_C_LOAD_UKER_G(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z30,p0,p0,x9,x7,x8,x16)
+GEMM_C_FMLA_UKER(z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,p0,z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z31)
 "                                                 \n\t"
 " BETA_ZERO_G:                                    \n\t"
-"                                                 \n\t"
-GEMM_C_STORE_UKER_G(z20,z22,z24,z26,z28,z21,z23,z25,z27,z29,z30,p0,p0,x5,x7,x8,x16)
 GEMM_C_STORE_UKER_G(z0,z2,z4,z6,z8,z1,z3,z5,z7,z9,z30,p0,p0,x5,x7,x8,x16)
+GEMM_C_STORE_UKER_G(z10,z12,z14,z16,z18,z11,z13,z15,z17,z19,z30,p0,p0,x5,x7,x8,x16)
 "                                                 \n\t"
 " END_WRITE_MEM:                                  \n\t"
 " b               END_EXEC                        \n\t"